A2A Registry: Complete Guide to Agent2Agent Protocol Registries in 2026
An A2A registry indexes AI agents that support Google's Agent2Agent protocol. This guide covers how A2A registries work, which registries support A2A, how to register your agent, and how Agent Cards drive discovery.

Laurent Yew
Founder
What Is an A2A Registry?
An A2A registry is a directory that indexes AI agents supporting Google's Agent2Agent (A2A) protocol. Each A2A registry lets developers discover, evaluate, and connect to agents that can communicate peer-to-peer using Agent Cards — structured JSON-LD documents describing an agent's capabilities, endpoint, and authentication requirements. Of the 15 registries indexed by AgentResourceDB, 6 currently support the A2A protocol, making A2A the second most broadly supported agent protocol after MCP.
If you're building multi-agent systems or deploying agents that need to collaborate with other agents, an A2A registry is where you list your agent so others can find and delegate tasks to it. This guide covers how A2A registries work, which registries support A2A, how to register your agent, and best practices for maximizing discoverability.
How A2A Registries Work
An A2A registry operates on a simple principle: agents publish Agent Cards, and other agents search those cards to find collaborators. When Agent A needs help with a subtask, it queries the registry for agents with matching capabilities, retrieves their Agent Cards, and initiates a direct peer-to-peer connection using the A2A protocol.
The Agent Card is the atomic unit of an A2A registry. It's a JSON-LD document hosted at a well-known URL (typically /.well-known/agent.json) that contains: the agent's name and description, a list of capabilities (what the agent can do), the endpoint URL where the agent receives A2A requests, authentication scheme (API key, OAuth, none), and optional metadata like version, owner, and rate limits.
When you register your agent on an A2A registry, the registry crawls your Agent Card endpoint, validates the JSON-LD structure, and indexes the capabilities. Other agents can then discover your agent by searching for specific capabilities like 'web-search' or 'code-execution' within the registry.
A2A-Compatible Registries in 2026
Not all AI agent registries support the A2A protocol. As of July 2026, 6 of the 15 registries indexed by AgentResourceDB accept A2A-protocol agents. Here's how they compare.
| Registry | A2A Support | Agent Cards | Listing Cost | Best For |
|---|---|---|---|---|
| Google AI Hub | Native | JSON-LD required | Free | Google ecosystem integration |
| LangChain Hub | Native | JSON-LD required | Free | LangChain multi-agent workflows |
| Pinecone Marketplace | Native | JSON-LD required | Free | Vector-search agents |
| CrewAI Registry | Supported | JSON-LD optional | Free | CrewAI team-based agents |
| AutoGPT Store | Supported | JSON-LD optional | Free | Autonomous goal-driven agents |
| AWS Bedrock | Supported | JSON-LD optional | AWS account | Enterprise AWS deployments |
// INFO
AgentResourceDB aggregates all 6 A2A-compatible registries into a single searchable index. Instead of checking each registry individually, search our registry and filter by A2A protocol to see every A2A-listed agent across all registries.
How to Register Your Agent on an A2A Registry
Registering your agent on an A2A registry takes four steps. The process is similar across all registries that support A2A, with minor differences in submission requirements.
- Host your Agent Card at a public HTTPS URL (typically https://yourdomain.com/.well-known/agent.json). The card must be valid JSON-LD with required fields: name, description, capabilities, endpoint, and authentication.
- Implement the A2A protocol at your endpoint. Your agent must accept HTTP POST requests with JSON-RPC payloads, respond to task delegation requests, and return structured results.
- Submit your Agent Card URL to the registry. Google AI Hub and LangChain Hub crawl and validate the card automatically. Other registries may require manual review.
- Verify your listing. After submission, search the registry for your agent name. AgentResourceDB indexes new A2A listings within 24-48 hours and assigns a trust score based on protocol compliance, uptime, and community signals.
A2A Registry vs MCP Registry: What's the Difference?
A common question is whether you should list your agent on an A2A registry, an MCP registry, or both. The answer depends on what your agent does. An A2A registry lists agents that collaborate with other agents. An MCP registry lists agents that connect to external tools and data sources. They serve different discovery patterns.
| Dimension | A2A Registry | MCP Registry |
|---|---|---|
| Discovery pattern | Agent finds agent | Agent finds tool |
| Protocol | Agent2Agent (Google) | Model Context Protocol (Anthropic) |
| Card format | Agent Card (JSON-LD) | MCP server manifest |
| Registries supporting it | 6 of 15 | 10 of 15 |
| Primary use case | Multi-agent collaboration | Tool integration |
| Can list together? | Yes — agents supporting both protocols appear in both registry types | Yes |
Most production agents should be listed on both. An agent that supports MCP for tool integration and A2A for agent collaboration appears in 15 of 15 registries when searching across AgentResourceDB — maximizing discoverability.
Agent Cards: The Key to A2A Discovery
Your Agent Card is the single most important factor in how discoverable your agent is on an A2A registry. A well-structured card with accurate capabilities gets matched to more search queries; a sparse or inaccurate card gets filtered out. Here's what makes an effective Agent Card.
- Use specific capability names. 'web-search' and 'pdf-extraction' are better than 'research' — specific capabilities match more targeted agent-to-agent queries.
- Include a clear description. Other agents use this to decide whether to delegate a task to you. Treat it like a technical spec, not a marketing blurb.
- Keep your endpoint live. A2A registries periodically re-crawl Agent Cards. If your endpoint is down during a crawl, your agent may be marked inactive.
- Version your card. Include a version field and update it when capabilities change. This helps registries detect stale listings and keeps discovery results current.
- Document authentication clearly. If your agent requires an API key, specify the header name. Ambiguous auth requirements cause delegation failures that lower your trust score.
Best Practices for A2A Registry Listing
Getting listed is just the start. How you maintain your A2A registry listing determines your long-term discoverability and trust score.
- Monitor uptime. AgentResourceDB's liveness checks ping your A2A endpoint every 15 minutes. Agents below 95% uptime receive a lowered trust score and may be filtered out of search results.
- Respond to delegation requests promptly. Agents that consistently time out on A2A task requests are ranked lower in registry search results.
- Update your Agent Card when capabilities change. If you add or remove a capability, update the card immediately so the registry reflects your current state.
- Register on multiple A2A registries. Listing on Google AI Hub, LangChain Hub, and Pinecone Marketplace triples your discovery surface. AgentResourceDB aggregates all three.
- Test your A2A implementation. Use the A2A protocol compliance checker to verify your agent correctly handles task delegation, status queries, and result retrieval before submitting to a registry.
// TIP
Browse the AgentResourceDB registry and filter by A2A protocol to see how top-ranked agents structure their Agent Cards. Look for agents with trust scores above 90 — their card structure and capability naming are proven patterns for A2A discoverability.
The Future of A2A Registries
A2A registry support is growing. When Google introduced the Agent2Agent protocol in 2025, only 2 registries supported it. By mid-2026, that number has grown to 6, and adoption is accelerating as multi-agent systems move from research to production. We expect at least 3 more registries to add A2A support by end of 2026.
AgentResourceDB will continue indexing every A2A-compatible registry as it emerges. Our trust score already factors in A2A protocol compliance, and we're adding real-time Agent Card validation to detect stale or broken listings automatically. Subscribe to our blog for updates as new registries add A2A support.
// Ready to explore?
Browse the full AgentResourceDB registry with 104,000+ AI agents across 15 registries.
Browse the Registry// Author

Laurent Yew
Founder
Laurent Yew is the founder of AgentResourceDB, where he leads the platform's vision of building a unified, trust-first discovery layer for the AI agent ecosystem. With over a decade of experience scaling AI and SaaS products, Laurent has dedicated his career to making complex developer infrastructure accessible, transparent, and reliable. He writes about agent registries, protocol interoperability, and the future of agent-to-agent collaboration, drawing from hands-on work building evaluation frameworks that help developers cut through the noise of 100,000+ agents. Through AgentResourceDB, he is committed to establishing the trust standards the industry needs as AI agents move from experimentation to production.
// Frequently Asked Questions
What is an A2A registry?
An A2A registry is a directory that indexes AI agents supporting Google's Agent2Agent (A2A) protocol. Agents publish Agent Cards (JSON-LD documents) describing their capabilities, and other agents search the registry to discover and delegate tasks to compatible collaborators. Six of the 15 registries indexed by AgentResourceDB currently support A2A.
How many registries support the A2A protocol?
As of July 2026, 6 of the 15 registries indexed by AgentResourceDB support the A2A protocol: Google AI Hub, LangChain Hub, Pinecone Marketplace, CrewAI Registry, AutoGPT Store, and AWS Bedrock. MCP has broader support at 10 of 15 registries.
How do I register my agent on an A2A registry?
To register on an A2A registry, host a valid Agent Card (JSON-LD) at a public HTTPS URL, implement the A2A protocol at your endpoint, submit your Agent Card URL to the registry, and verify your listing. Google AI Hub and LangChain Hub validate automatically; other registries may require manual review.
What is the difference between an A2A registry and an MCP registry?
An A2A registry lists agents that collaborate with other agents (agent-to-agent discovery), while an MCP registry lists agents that connect to external tools and data sources (agent-to-tool discovery). Agents supporting both protocols can be listed on both registry types, appearing in 15 of 15 registries on AgentResourceDB.
What is an Agent Card in A2A?
An Agent Card is a JSON-LD document hosted at a well-known URL (typically /.well-known/agent.json) that describes an AI agent's name, capabilities, endpoint, and authentication requirements. A2A registries crawl and index Agent Cards so other agents can discover and delegate tasks to compatible collaborators.